... Yes and no. We use 8-bit heightmaps all the time. If you change the scale (we use 1 unit = 1/4th of a meter IIRC and each pixel represents 1 meter) it...
383
jimmyp@...
Jun 1, 2001 8:24 pm
... errors ... variables ... I think that what you say here isn't entirely correct(at least it doesn't work for me).Assuming eT,eT0 and eT1(eT values for...
384
Ben Discoe
ben@...
Jun 4, 2001 7:07 am
Those of you wishing to model Australia may find the following new imagery source interesting. It is a extremely large dataset of aerial images and related...
385
Mark Duchaineau
duchaine@...
Jun 4, 2001 6:22 pm
Jimmy, Just a minor clarification: when you store eT, you don't actually change eT, you just say eT_array[idx]=(eT+1)>>1. This means that you can return eT,...
386
Garry Keltie
garry.keltie@...
Jun 5, 2001 12:04 am
This is interesting and would represent a major breakthrough in licencing agreements with Land Information. Can Peter please confirm for the group that this...
387
jimmyp@...
Jun 6, 2001 12:10 pm
Since everything seemed to be running smoothly with split-only I tried moving on to split-merge.I had all the functions needed allready since I previously had...
388
Lucas Ackerman
ackerman@...
Jun 6, 2001 5:28 pm
hi Jimmy. to update a queue, you generally want to be transferring all queue entries into a seperate queue, so you don't do multiple updates on a tri. since...
389
jimmyp@...
Jun 8, 2001 10:35 am
... all ... updates on a ... lists, ... after ... Hi Lucas, I too use 2 pointer arrays.Something like qnode_t *qs[QS_BUCKETS] and qnode_t *qm[QM_BUCKETS],where...
390
Lucas Ackerman
ackerman@...
Jun 9, 2001 12:49 am
... yes, 2 arrays per queue. transferring one to the other is potentially much faster than doing an extra tree tranversal (which could touch twice as many...
391
jimmyp@...
Jun 9, 2001 11:17 am
... tri's ... depth-first ... because you ... Is that what Mark describes in the suggested implementation phases of the roam hoamepage?(at the end) ... fairly ...
392
Lucas Ackerman
ackerman@...
Jun 11, 2001 6:12 pm
... no. the final bit of that page is about how to generate actual strips. ... this is incorrect. it works for 2d, where 'r' would be the distance to each...
393
Mark Duchaineau
duchaine@...
Jun 11, 2001 10:19 pm
Hi Jimmy, Lucas kindly answered many of your questions. I'll take the one on the linear-order error estimate. First off, the basic idea is to look at the...
394
Mark Duchaineau
duchaine@...
Jun 11, 2001 10:28 pm
Minor correction: ... I just realized this is silly. Don't do this. You want an estimate on how far any point is on the screen from where it should be. The...
395
jimmyp@...
Jun 12, 2001 12:15 pm
Hi Lucas, ... of ... strips. ... Uhh,is there any info on this anywhere(message-list thread etc...)? ... to ... you can ... every time Average of 3 verts?As...
396
jimmyp@...
Jun 12, 2001 12:40 pm
Hi Mark, glad to be able to be of some help for a change(regarding that bug), ... going ... If I don't understand this I'll just have to keep asking you every ...
397
jimmyp@...
Jun 15, 2001 1:26 am
... Hello again Lucas, I have tried both algorothms you suggested. You wrote: you can ... every time ... bound. To test the performance of each algorithm...
398
vcruz@...
Jun 15, 2001 9:41 am
... Hi Jimy, (...) ... the ... do ... if the viewing transform includes shearing and scaling w1 will the z of the point in the camera space (without scale and...
399
vcruz@...
Jun 15, 2001 9:44 am
Hi, I realized that cut/paste is evil :) In R matrix, U is the right vector, V the up vector, and W the direction vector....
400
Dimitris Papavasiliou
jimmyp@...
Jun 15, 2001 4:12 pm
Hi there Cruz, ... Hmmm,are you sure about this?I only R currently,(or something that looks like it at least).I take it that R does rotation and...
401
vcruz@...
Jun 15, 2001 4:51 pm
... the z ... looks like it at least).I take it that R does rotation and transforamtion?And V does something related to frustum planes?What does it do...
402
Lucas Ackerman
ackerman@...
Jun 15, 2001 7:06 pm
hi Jimmy. I don't recommend disabling incrimental (parent->child) frust flag updating, because while yes, you will get a more thorough measure of how fast each...
403
Dimitris Papavasiliou
jimmyp@...
Jun 15, 2001 10:59 pm
Hello again Vincent, ... Thanks for the link.I wish I had it when I was struggling with the camera model. ... I also assume you assume corrsctly but you didn't...
404
Dimitris Papavasiliou
jimmyp@...
Jun 15, 2001 10:59 pm
Lucas, we seem to be getting somewhere here, ... I rtied both w/ and w/o disbaled incrimental updating.I disabled it to see the 'raw' speed of each method...
405
vcruz@...
Jun 16, 2001 4:20 pm
... Hy Dimitris, V is the viewing transformation matrix which tranform the coordinate from the camera space to the screen space. Take a look at : ...
406
Dimitris Papavasiliou
jimmyp@...
Jun 16, 2001 4:55 pm
... I thought I was suppsed to use camera-space coords.So the matrix M trans- forms to device coords? ... Do you mean you use the float priority you calculate...
407
vcruz@...
Jun 18, 2001 8:25 am
... coordinate ... trans- ... Yes. But if the point is in the view frustum, all its coordinates are in the range [-1.0,1.0] ... (minMergePriority < ... someone...
408
Lucas Ackerman
ackerman@...
Jun 18, 2001 7:16 pm
... ah, this is perhaps our prime miscommunication: when I say 'store' anything, I mean ONCE at boot time as a pre-compute step, not everytime the obj is...
409
Maltez
maltez@...
Jun 18, 2001 10:10 pm
Hi I used a heapsort queue in my first implementation of ROAM. It worked fine : i couldn't time the overhead! In my current re-implentation, i use an unsorted...
410
Dimitris Papavasiliou
jimmyp@...
Jun 19, 2001 1:11 pm
... Precompute for every tri?Like we do for nested bounds?That should take quite a few megs of mem!No wonder you like method B) better:) ... I see.On a...
411
Dimitris Papavasiliou
jimmyp@...
Jun 19, 2001 1:12 pm
Hi Maltez, ... Well you're supposed to split the max split pri triangle and merge the min merge pri so you should only need a pointer to these so generally...
